Nowadays, with the demand to reflect the real world, we have a number of imprecise stored business data warehouses. The precise data classification cannot solve all the requirements. Thus, the fuzzy decision tree classification problem is important for the fuzzy data mining problem. The fuzzy decision classification based on the fuzzy set theory has some limitations derived from its innerself. The hedge algebra with many advantages has become a really useful tool for solving the fuzzy decision tree classification. However, the sample data homogenising process based on the quantitative methods of the hedge algebra still has some restrictions because of errors evolved and the resulting tree is not truly flexible. So, the fuzzy decision tree obtained is not always highly predictable. In this paper, using fuzziness intervals matching with hedge algebra, the authors proposed an inductive learning method "HAC4.5 fuzzy decision tree" to obtain a fuzzy decision tree with high predictability.